Blurring Borders: the Significance of Humantiarian Safe Spaces
Mothers of Srebrenica, an association of relatives of thousands of Muslims killed by
Bosnian Serbs in 1995, said they would take legal action against United Nations officials
for failing to stop the killing. In a report last month, the United Nations blamed itself for
failing to prevent the deaths.(The New York Times, December 2, 1999).1
A group of survivors of the 1995 massacre of up to 8,000 Bosnian Muslims in Srebrenica
tried to file a complaint at the international war crimes tribunal in The Hague, charging
Secretary General Kofi Annan and other officials with partial responsibility for failing to
protect the town. Fred Eckhard, Mr. Annan‘s spokesman, said the tribunal does not allow
individuals to bring cases, saying that right rests solely with its chief prosecutor, Carla del
Ponte (The New York Times, February 5, 2000).
Srebrenica, a mainly Muslim town in eastern Bosnia-Herzegovina (BiH), was designated a ‗safe haven‘
by the UN Security Council in a resolution passed on April 16, 1993. The Security Council added five
more ‗safe areas‘, all with predominantly Muslim populations, in a second resolution on May 6, 1993. A
third resolution, passed June 4, 1993, authorized air strikes to protect the six areas‘ UN defenders.
Despite these measures, things went seriously awry in Srebrenica in July 1995. Led by Ratko Mladic,
Bosnian Serb forces entered the town where some 7-8,000 Muslims, most of whom were men or boys,
were murdered. ―Srebrenica has fallen…. The worst scenario that we feared has come true, and the Serbs
are sweeping the pocket clean‖ (Kris Jankowski of the United Nations High Commissioner for Refugees
(UNHCR), cited in Power, 1995). Unlike the safe havens of Sarajevo, Gorazde, and Bihac which had
large Bosnian army forces, Srebrenica was mostly demilitarized. Consequently, Serbs were able to take
In November 1999, UN Secretary-General, Kofi Annan, released a highly self-critical report about the so-called safe area of
Srebrenica: ―Through error, misjudegment and the inability to recognize the scope of evil confronting us we failed to do our
part to save the people of Srebrenica from the Serb campaign of mass murder…. These failings were in part rooted in a
philosophy of netrality and nonviolence wholly unsuited to the conflict in Bosnia…. In particular, the report makes clear the
inadequacy of the entire approach of the United Nations to the Serb campaign of ethnic cleansing and mass murder,
culminating at Srebrenica‖ (UN official cited in Crossette, 1999). The 155 page self-indictment is the basis upon which the
Mothers of Srebrenica group launched their legal action against the UN.
over the town with little resistance except for the small battalion of (110-439) Dutch peacekeepers who
left the town when their air cover did not arrive, and they realized the futility of their presence (HINA
news agency, 1998).
Srebrenica is just one experiment of creating safe spaces in war zones that occurred during the 1990s.
The idea was first tried in Northern Iraq in 1991, with UN Operation Provide Comfort. Some 400,000
Kurds fleeing the violence of Saddam Hussein‘s forces approached the Turkish border, where they were
not welcome. In an unprecedented move, the UN Security Council passed resolution 688 creating a ‗safe
haven‘ in Northern Iraq. Protected by international peacekeepers on the ground, and a ‗no fly zone‘ in
the air space above, Operation Provide Comfort was — in hindsight — the most effective of the safe
haven attempts over the past decade, after which ―the standard of safety in such zones steadily declined‖
(Frelick, 1999: 24). From Iraq to Haiti to Rwanda, different safe areas have been deployed with disparate
results and to different ends. If safe spaces are more ―rhetoric than reality‖ (ibid.: 25), then why have
they been used so often by international bodies mediating the consequences of war? Why are some safe
areas more effective than others? What political, legal, and geographical antecedents generate such
This paper analyzes the meanings, outcomes, and geographies of three distinct safe spaces in Somalia,
Bosnia-Herzegovina, and Sri Lanka. Each of these safe spaces is situated within a unique context of war,
and has its own discursive antecedents and pre-war identity. I contend that the safe areas have specific
antecedents which shape their effectiveness and outcomes. The humanitarian discourse of ‗preventive
protection‘ and ‗the right to remain‘ has generated many, but not all, of the safe spaces within conflict
areas in the 1990s. This new language of humanitarian assistance has emerged as a highly spatialized
project of reorienting human displacement closer to home. Crossing an international border and claiming
asylum as a refugee is no longer the only way to receive assistance and/or protection from the
international humanitarian regime. The meaning of ‗refugee‘ has changed dramatically since the end of
the Cold War, and its value in geopolitical terms has declined. Where refugees once provided proof of
Western superiority in its rivalry and tension with the USSR, they are increasingly a product of civil
conflicts, and yet they remain a global concern if their forced migration crosses international borders.
Political support for hosting refugees has reached its limits among many governments in countries
receiving asylum seekers. As Kathleen Newland (1999: 17) notes, ―a determination was made to bring
safety to people rather than people to safety, by force if necessary.‖ The strategy of helping people
internally displaced within their home countries is both a political and practical one which has met with
mixed results. One goal of this paper, then, is to trace the way in which these safe areas were established,
ascertain their relative safety, and determine why some are more effective than others. Whether the safe
areas are predicated on legal, political, or geographical grounds makes all the difference.
There is madness in my method. The analysis presented is not strictly comparative, that is, the
approaches taken to glean insights about each of these sites were distinct. My field visits to safe areas in
Somalia, Sri Lanka, and Bosnia took place over a period of seven years, under different circumstances of
war and peace, and in different research capacities. My access to these areas varied considerably: in
Somalia, I first worked for UNHCR to implement the objectives of the Cross-Border Operation based on
the establishment of ‗preventive zone‘ in the southern region of the country, and then returned to the
region in 1994-95 as a researcher. In 1999, my fieldwork in Bosnia was circumscribed by the aftermath
of the NATO bombing. As someone from a NATO country, and based on travel warnings issued by such
countries to their citizens, I did not visit Srebrenica, located in what is now the predominantly Serb,
Republic of Srpska. I focused my fieldwork instead in Sarajevo and Gorazde. In Sri Lanka, my work on
a related research project took me all over the country, including to Madhu, the location of a very
distinct and arguably successful safe area, one that underwent considerable upheaval in May 1999 just
before my visit. The information analyzed in the paper derives from field notes made during visits to
each of these locations based on interviews with local staff in UN agencies, non-governmental
organizations (NGOs), and other contacts. I also analyze newspaper reports, evaluations of operations,
and a number of secondary sources.
New Spaces, new strategies
The end of the Cold War signalled the cessation of superpower conflict. Hopes for greater peace on a
global scale, however, have been dashed by the rise of ethnic nationalism, civil conflict, and secessionist
claims. The 1990s witnessed a respatialization of responses to crises of human displacement, preferring
solutions that addressed the problem as close to home as possible (Hyndman, 1999). The international
refugee regime is increasingly referred to as the international humanitarian regime, in which legal
protection has been largely supplanted by material assistance provided in politically circumscribed
humanitarian zones. The major players in this regime, namely donor governments of the most
industrialized countries, are shifting their emphasis from ‗the right to leave‘ one‘s country in the face of
persecution or violence to the ‗the right to remain.‘ Displaced people are encouraged to stay within their
countries of origin by providing assistance to them there, rather than having them seek protection
through asylum across an international border. In some of these situations, designated safe areas were
created within war zones to provide humanitarian assistance and to protect civilians at risk.
Early in the decade, law professor James Hathaway (1994) called the right to remain, the ―right to be
toast‖ because of the inherent dangers of the strategy. Some policy analysts say the concept violates the
right to leave one‘s country outlined in the UN Declaration of Human Rights. By the end of the
decade‘s experiments in safe areas, experts in humanitarian law and operations articulated incisive
questions about the viability of such protected zones, yet they had few answers. Mr. Sergio Vieira de
Mello, Under-Secretary General for Humanitarian Affairs at the UN Office for the Coordination of
Humanitarian Affairs (OCHA) noted that the idea of protected areas raises as many questions as it
appears to resolve.
Such questions include:
Whether the idea of establishing a protected area based on consent is realistic?
Whether the establishment of a protected area using international force can be effective?
Whether the creation of such areas might contribute (unwittingly) to ethnic cleansing?
To what extent the creation of a protected area undermines the right to asylum?
To what extent might the creation of such an area permanently influence international borders?
To what extent does the construction of such an area affect the treatment of civilians outside the area?
(OCHA, 1999: 2)
What is most interesting about these questions is that they could only be asked in hindsight, after well-
intentioned experiments in deploying safe spaces had been tried with mixed results. If these questions
could have been posed and answered at the beginning of the 1990s, patterns of humanitarian response
may well have looked very different.
The shift in discourse from ‗the right to leave‘ and ‗the right to asylum‘ to ‗the right to remain‘ has
generated one set of safe areas, a set that has come into being as a result of UN Security Council
resolutions. ‗Preventive protection‘ designates a new set of safe spaces and management practices for
people out of place, literally those forced to flee their homes. While the establishment of 'safe havens' for
Iraqi Kurds in Northern Iraq in 1991 is said to mark this turning point, 'zones of tranquillity' for returning
Afghan refugees, 'open relief centers' for would-be Sri Lankan refugees, and 'safe corridors' to Muslim
enclaves in Bosnia are also expression of this respatialization of humanitarian response. Humanitarian
law has long defined another set of safe spaces, including hospitals for combatants and agreed upon
treatment for prisoners of war. More traditional safe spaces, namely refugee camps, have also played an
important role in providing sanctuary for displaced person, though even the safety of camps has come
under question since the Rwanda crisis.2
UN perspectives on safe areas
The consultation with experts organized by OCHA and Harvard Center for Population and Development
Studies in February 1999 generated important insights and new terminology around the concept of safe
spaces. From the outset, the geographical and historical contingency of protection was identified: ―The
concept of protection was recognized to be sensitive to elements of place (i.e., terrain, climate,
geographic location) size (i.e., population, spatial area), time (i.e., duration, stage of conflict) and
overlapping regulatory regimes‖ (OHCA, 1999: 5). Despite this important introductory observation and
subsequent comments, little genuine analysis of the historically constituted meanings and stories of safe
areas was undertaken. Rather, a familiar debate between advocates of humanitarian law versus those of
humanitarian operations authorized by the UN Security Council resolutions ensued. This tension
between legal and more political approaches to complex emergencies is nothing new.
The very meaning of humanitarian protection was debated at length at the consultation, with participants
citing distinct notions of prevention in humanitarian law and in practice, as authorized by the UN
(OCHA, 1999). Preventive protection defined in relation to international humanitarian law is contractual
and based on consent by warring parties which agree to render designated areas of medical treatment ‗off
limits.‘ Preventive protection in the operational sense, in contrast, has generally referred to the
designation of safe spaces by UN authorities enforced by military power, usually in the form of
peacekeepers. The distinction between consensual ‗protected zones‘ based in international humanitarian
Refugees fleeing Rwanda to (then) Zaire were accompanied by a number of perpetrators of the genocide against Tutsi in
Rwanda. Ample evidence that these perpetrators held refugees in the camps hostage to their wishes has generated much soul-
law and ‗safe areas‘ specified by the UN Security Council, but not necessarily agreed to by warring
factions, is a crucial one. The participants determined that consensual safe areas should be called
‗humanitarian zones‘ and the UN-designated sites ‗security zones‘, a terminology that suggests a
distinction between legal and politically-driven safe areas. Yet, as before, there is little recognition of
local patterns of protection or consent. Protection in both the legal and political sense is assumed to be
the domain of international actors and instruments, as evinced in the experts‘ ―[r]ecognition of the need
to ‗manage‘ consent and to try to preempt its erosion or withdrawal‖ (ibid.: 18).3 A more comprehensive
treatment of geographical variation and contingencies of time and place, however, could have generated
valuable and innovative insights.4
This spatially blind approach to human protection underscores the absence of a geography of safety in
the context of war and humanitarian response. The conditions for a geography, or geopolitics, of human
protection — which attends to the cultural politics, geopolitics, and history of place — emerge between
these legal and political approaches. Human security during war cannot rely solely upon contractual legal
obligations at the international level, nor can the security of states prevail over the security of people if
genuine protection is the aim. Human security relies upon a finer scale of analysis, that of persons not
simply states, and a broader sense of what constitutes security (DFAIT, 1999). The historical relation of
particular groups to specific areas, whether they are designated ‗safe‘ by international authorities or not,
is not even part of the puzzle experts addressed when they took an inventory of UN protected areas.5
searching among humanitarian organizations that operated in Zaire (Hyndman, 2000).
Concern was raised that the methodology employed by the group placed humanitarian actors ―in the compromising position
of assuming the role of an obligor under human rights law, a role which (it was emphasized) states alone should play….
humanitarian actors could adopt an analytical metholodgy that drew from human rights discourse without undermining the
legal obligations of states or purporting to assume the role of obligors‖ (OCHA, 1999: 13).
The significance of timing, geography, and the political dimensions of humanitarian activities was acknowledged, including
a geopolitical analysis of escape routes and possible catchment areas for Albanian Kosovars (OCHA, 1999:14-15).
The expert group did identify as a topic for further studey the ―need to explore ad hoc consensual solutions outside the
traditional normative framework, e.g., Open Relief Centers in Sri Lanka‖ (OCHA, 1999: 18). The relative success of safe
Of humanitarian zones and security zones articulated, the more controversial has been the deployment of
the security zone, as in Bosnia. Ironically, UN authorities achieved a negotiated agreement with the
Bosnian Serbs on Srebrenica as a safe city, unlike the safe cities of Sarajevo and Gorazde, yet this form
of consent meant little: Srebrenica was attacked and much of its population massacred while Sarajevo
and Gorazde were safeguarded. The experts agreed that such a tactic should be a strategy of last resort in
locations where no trust or respect for international humanitarian law can be established. Security zones
are not economically feasible, generate unavoidable dependency on international assistance, and have the
potential to create ―a largely vacant landscape dotted with small, unsustainable protected areas‖ (ibid.:
8). Concern was voiced that protected areas were represented as ‗empty spaces‘ that failed to take
account of the socio-political relations and organizations of civil society in situ . Such imperial
cartographies have long served to obfuscate the presence of indigenous peoples and the political
importance of their struggles (Sparke, 1995; Harley, 1992).
I move now to Bosnia where the experiment in safe spaces went seriously awry. Having introduced
several of the cities and towns designated as safe, this section briefly traces the terrain of conflict and
displacement, the politicization of the humanitarian project, and the violation of protection provided by
the safe havens of Zepa and Srebrenica in particular.
Between 1992 and 1995, brutal and systematic campaigns of ethnic cleansing in Bosnia displaced more
than half the population (Cutts, 1999). By the end of the war, almost a million people were refugees in
areas in Sri Lanka as compared to Somalia and Bosnia is, I will argue, based on the recognition of local meanings and politics
of the place in which the protected areas were established.
other countries while 1.3 million people, out of the area‘s pre-war population of 4.3 million, were
internally displaced. Subsequently, hundreds of thousands of people became dependent on humanitarian
aid, especially in the designated safe havens where rapid urbanization to these cities and towns left large
parts of the countryside abandoned.6
The causes of the war are much more difficult to recite, and a comprehensive treatment of its dynamics
is precluded here (see Owen, 1995; Glenny, 1992). However, claims to the territory of Bosnia-
Herzegovina took on a renewed urgency after the European Union recognized the secession of the
republics of Slovenia and Croatia from the state of Yugoslavia in 1991. Bosnia-Herzegovina had little
choice but to declare its own independence from Belgrade.7 It would have otherwise become a largely
non-Serbian republic dominated by an increasingly Serb-dominated country, or it would have to accept
the division of Bosnia between Serbs and Croats, as proposed by their respective leaders.8 Bosnia‘s
ethnic composition before the war was approximately 44% Muslim, 31% Serb, 17% Croat, 6% Yugoslav
with a small residual population (Campbell, 1999).9 Bosnia‘s declaration of independence was, of
Four years after the war has ended, 1.2 million people remain either internally displaced or live as refuees abroad, unable or
unwilling to return to their prewar homes (New York Times, 2000).
Bosnia is a cultural and political geographer‘s dream; in Misha Glenny‘s words:
[I]t is through the middle of Bosnia that East meets West; Islam meets Christianity; the Catholic eyes the Orthodox
across the Neretva, the line of the Great Schism; Bosnia divided the great empires of Vienna and Constantinople;
Bosnia was perhaps the only tue reflection of Yugoslavia…. Above all, history springs from the four rivers, the Una,
the Sava, the Drina and the Neretva which, with the exception of the latter, outline what appear almost natural
borders of this artificial and yet eminently durable political unit (Glenny, 1992: 161).
UN Security Council resolution 755 on May 20, 1992 recommended that Bosnia and Herzegovina be admitted to
membership of the United Nations.
Campbell notes that few academic discussions mention ethnic breakdown. My own reading of this is that academic analyses
tend to situate the roots of conflict in nationalism, whereas the media have often rendered the problem as ‗ethnic.‘ In a 1992
study commissioned by the Bosnian presidency, over 10% (56,473) of Sarajevo‘s population still called themselves
Yugoslavs, despite the violence. Bosnia had the highest percentage of those who self-identified as Yugoslavs in the national
census. Glenny (1992: 142) argues that the move from people identifying as ‗Bosnian‘ to identifying as Bosnian Serb,
Bosnian Croat, or Bosnian Muslim was the initial step to war in the republic. Glenny also notes that the bright-green eyes
found among all three national groups challenge any theories of racial purity.
Like the Muslims of Sri Lanka, the Slav Muslims of Bosnia are one of the few nations in the world who are nominally
identified by their religion and not their language or ethnicity. Bosnian Muslims were recognized as a constituent nation, and
not just a minority, within Yugoslavia in 1971, and enshrined in the constitution in 1974 under Tito‘s rule (Glenny, 1992).
course, contested by the Serb-dominated army of Yugoslavia under the control of Slobodan Milosevic
and by the president of Croatia, Franjo Tudman.10 Croats and Serbs continued to fight by proxy in and
over Bosnia-Hercegovina. Misha Glenny‘s clairvoyant 1992 book about violence and tensions in this
region predicts the war in Bosnia-Herzegovina and identifies its roots:
The essential problem of a Yugoslav state lies in the numerical and political dominance of Serbs over Croats; the
essential problem of a Croatian state lies in the numerical and political dominance of Croats over Serbs. In order to
secure peace in the Balkans, this conundrum must be solved along with two others: the constitution of Bosnia-
Hercegovina and in Serbia, the political status of the Moslem, Albanian and Hungarian minorities in the Sandzak,
Kosovo and Vojvodina, respectively (Glenny, 1992: 100).
According to Glenny, the end of Bosnia began the weekend of March 1, 1992 when the European Union
announced that its independence would be recognized. Between May 1992 and November 1995, the UN
Security Council passed 46 resolutions concerning the situation in Bosnia-Hercegovina (Cutts, 1999). It
provided both humanitarian and peacekeeping missions to Bosnia to ensure the delivery and distribution
of humanitarian assistance. The UN Protection Force, UNPROFOR, was assigned as the peacekeeping
mission to Bosnia, while UNHCR was designated the lead agency for humanitarian operations. The was
the first time that UNHCR had worked directly in a conflict zone (Cunliffe and Pugh, 1997).11
As noted at the outset, two UN Security Council resolutions (824 and 836) created the basis for the ‗safe
areas‘ of Srebrenica, Sarajevo, Tuzla, Zepa, Gorazde, and Bihac. All of these centres hosted a majority
Bosnian Muslim population, but cities like Tuzla and Sarajevo were renowned for their multi-ethnic
character. Bihac, a mainly Muslim town located in the northwest corner of Bosnia, was surrounded by
Serb-held territory. But the selection of Bihac as a protected area belies an important political history.
The Bihac-Cazin region has long featured a Muslim enclave amid a largely Serb population which
straddles both Bosnian and Croatian territory. Prior to the war, an alliance between Muslims and Serbs
Over 60% of Yugoslavia‘s military industries were based in Bosnia, 60% of which were located in Corat or Muslim regions
(Glenny, 1992: 150).
had existed here since World War II. In fact, Bihac was one of the safer UN protected areas during the
war, despite Bihac becoming the capital of Bosnian Serb nationalism.
Sarajevo (show slides) verbal text
— Bosnian nationalism is high (stickers on every car, many supplied by EU); it is a Bosnian Muslim
nationalism, as one interviewee informed me, as the Muslims are the only one not to have their own
country (this can be seen in the green coloured signs used for buildings). Josep (UNHCR) estimates that
post-war Sarajevo is now 80% Muslim, 5% Croat, and 15% Serb. This issue of Muslim dominance was
echoed by another source, Rajeeb Khanal (Nepalese physician, trained in USSR, and project officer for a
Japanese NGO called JEN), who voiced concern that the number two leader in Bosnia is a cleric/scholar
of Islam and a potential problem. Rajeeb finds the fundamentalist tack of this Islamicization trend
disturbing; it has adversely affected the position of women in Bosnian society and national sentiment is
brewing just beneath the surface. He feels that this trend may be offset by strong economic development
which trickles down so that people begin to feel better off. Rajeeb compared Bosnia today with
Yugoslavia in 1985, when people were free leave the country and work as migrants in Germany, but
often they did not because they could earn a decent salary at home (1500DM/month compared with 200-
— Despite the end of the war, many people are leaving Sarajevo. Average wages are low: 300-
400DM/month ($150-210 US or $250 –310 CDN); Selma‘s mother is the senior physician in a clinic and
gets 400 DM per month. This is nothing given that food for 4 is 435 DM/month, rent is high (higher
because of all the internationals around, 2000DM/month not uncommon). Compared to a year ago, the
During the war, more than fifty humanitarian personnel and over eighty UNPROFOR soldiers were killed in war-related
incidents (Cutts, 1999).
political situation is more settled but the economic situation is more precarious. Many people are leaving
for this reason. Selma herself is considering accepting a two year scholarship abroad to finish her
studies. Nada, another informant, listed the following issues as most important to women: poverty,
unemployment, low levels of education, and illiteracy. Nada explained that the war had generated a lot
of rural-urban migration, which has had the effect of generating less cosmopolitan attitudes. Just as the
seeds of nationalism were most virulent in the rural areas, women are more likely to represent their
national interests than ever before.
— Religious groups are active here (I met two young Americans who were working for a church in the
US, setting up in Macedonia and Bosnia). Selma says that people who join certain churches can get visas
to leave within a week.
On the bus to Gorazde, one encounters a sign about 10 kilometres outside Sarajevo which welcomes you
to Republika Srpska, if you are a foreigner, and ostensibly warns you, if you are not Serbian. A house
along the way displays the name ‗Dayton.‘ Needless to say, much of the post-war reconstruction of such
houses is based on international aid that came into Bosnia after the Dayton Accord was signed in 1995.
One leaves Republika Srpska before entering the small city of Gorazde, which straddles the Drina River
and sits amidst the small mountains which dot the area. One informant explained that the city has a post-
war population significantly less than that before the war. Many of the Serbs did not return to Gorazde,
one of the safe areas for Muslims during the war which today is a small pocket of the Croat-Muslim
Federation surrounded by the Serbian-dominated Republika Srpska on the new map of Bosnia-
Herzegovina. This is a place which now ―appears safe enough to leave a mountain bike unlocked, clean
enough to fish, and paved enough to roller blade‖ (field notes). The cafés and street life seem very lively,
but the scenery belies the sordid past when Muslims fled their homes in the countryside to find sanctuary
This town has normalized the least of all the former safe havens. I wouldn‘t go here, but even if I had, I
was told, ―there is nothing there.‖ The woman at the tourist office said she would not feel comfortable
on her own, and as a NATO passport holder, neither did I. Basically, local (Serb) authorities in
Srebrenica (which is now located in Republika Srpska) have refused to allow Bosniaks (Muslims) back
into Srebrenica, a town once 95% Muslim and 5% Serb. In June 1999, local Serb nationalists finally
allowed the majority Muslim town council, which was elected more than two years ago in absentia, to
take office (Rohde, 1999). Bowing to international pressure and local economic realities, the Serb
nationalists have nonetheless used endless negotiations, administrative bureaucracy, rock-throwing, and
other tactics to prevent the Muslim town councilors from meeting and to prevent Muslim refugees from
returning to their homes. The 12,000 Serbs who currently inhabit Srebrenica are mostly refugees from
Sarajevo, which ironically hosts most of the refugees who survived Srebrenica. Because of a four year
embargo by the US and EU nations, almost all international aid has bypassed Srebrenica in response to
On July 11, 1998 — exactly three years after the massacre at Srebrenica — a group of 30 women from
the town gathered in front of the office of the international high representative to Bosnia in Sarajevo to
hand him a list with the names of family members still filed as missing (HINA news agency, 1998).
Although some 2,200 bodies have been recovered so far, the remainder are missing. Without death
certificates for those likely killed in the massacre, surviving family members cannot file the documents
necessary to return to Srebrenica. The women demanded the documentation that Serbian authorities have
refused to provide, as this would mean eviction for a number of Serb refugees currently living in the
houses of former Muslim residents of Srebrenica.
In the broader national context of Bosnia, two critical sets of conditions served to destabilize the safe
cities of Bosnia. First, ―UNPROFOR was singularly unsuccessful at improving access for humanitarian
organizations to the government enclaves which were besieged by Bosnian Serb authorities‖ (Cutts,
1999). Mark Cutts, who was Head of the UNHCR office in Sarajevo for much of the war, explains that
this arrangement was the outcome of an arrangement in which UNPROFOR itself depended entirely on
authorization from the Bosnian Serb authorities to travel through its territory. Thus, permission to travel
to the safe areas in order to protect displaced civilians was sought from the very forces who threatened
their safety! When Bosnian Serbs cut off humanitarian access to Sarajevo, food and medical supplies fell
to dangerously low levels, despite the UN airlift.12 ―[A]ssistance was in fact provided on the basis of
accessibility rather than on the basis of needs (Cutts, 1999; emphasis in original).
Second, while UNPROFOR‘s peacekeeping position was less than optimal, UNHCR humanitarian
mission also encountered problems detrimental to the safe areas. As the lead civilian agency, UNHCR
became politicized by forces beyond its control. Technically an apolitical agency, UNHCR‘s role in the
conflict was increasingly seen as political rather than humanitarian as it negotiated the proportional
distribution of humanitarian goods and appeared, at times, to be in charge of protecting Bosnian
Muslims. ―This process [of politicization] was compounded in three ways: the absence of international
political will; the financial and political pressures imposed upon UNHCR‘s operational environment;
and the blurring of the humanitarian and military operations‖ (Cunliffe and Pugh, 1997: 137-38). Given
UNPROFOR‘s reliance on Serb officials for permission to travel, UNHCR‘s reliance on UNPROFOR
for aid convoys meant that it fell prey to similar problems All of these pressures ―have coincided with
emphasis upon ‗country of origin solutions‘ to eradicate the causes of refugee flight‖ (ibid.: 141-42). Not
only did UNHCR have limited direct representation in the UN protected areas, but the agency could not
ensure the protection of basic rights and had little control over the physical protection of civilians in the
region.13 The provision of humanitarian assistance substituted for protection. ―While UNHCR was able
to deliver large quantities of humanitarian supplies during the war, it was much less successful in
carrying out its protection mandate‖ (Cutts, 1999).
The ‗preventive zone‘ established in Somalia was less a UN Security Council invention than a political
panacea. The establishment of preventive zone inside Somalia was part of a larger mission, the Cross-
Border Operation (CBO), instigated late in 1992 and ―aimed at preventing new refugees and facilitating
repatriation‖ (Kirkby, 1997: 181). The CBO was in large part a response to the threats of Kenyan
President Daniel Arap Moi, whose reelection in December 1992 provided a political platform to
announce the Kenyan Government‘s plan to return Somalia refugees — by whatever means necessary —
to their country. UNHCR-sponsored refugee camps in Kenya at the time hosted hundreds of thousands
of refugees, but the UNHCR‘s work with refugees relied on the permission of the Kenyan Government.
UNHCR could not operate on the same scale within Kenya, and so sustained efforts to fund an
Between July 1992 and January 1996, UNHCR coordinated the longest running humanitarian airlift in history. It surpassed
the Berlin airlift in duration, providing more than 12,000 flights and medical evacuation for over 1,100 casualties of the war
Ethnic cleansing proved to be one of the biggest conundrums for UNHCR. Could they evacuate civilians who were under
threat but who had not yet fled? Or was that aiding and abetting the enemy by helping it to ethnically cleanse the areas under
alternative ‗preventive‘ path ensued. Thus, the preventive zone had a humanitarian purpose, but was
produced by political exigency.
In January 1991, Somalian President Siad Barre was ousted from power. By 1992, civil conflict had
become widespread and had induced famine in several parts of the country. Images of malnourished
Somalians appeared on televisions worldwide, winning public sympathy and government donations to
fund humanitarian efforts in both Somalia and Kenya. A series of refugee camps and temporary border
sites were established in Kenya to accommodate Somalians as they crossed the border in desperate
physical condition. Mortality rates soared in these ill-prepared makeshift camps until water quality,
sanitation conditions, and food supplies could be stabilized. Deaths were counted by the number of
shrouds — simple pieces of white cloth distributed on demand — and the number of bodies discernible
in graveyards located at the perimeter of the camps. Working in one of the Kenyan refugee camps at that
time, I witnessed the costs and corporeality of displacement on a human scale.
In December 1992, the US-led Operation Restore Hope landed in Somalia to save the country from
itself. Perhaps the most vivid testimony of prevention was the passing of UN Security Council
Resolution 794 which authorized a Unified Task Force (UNITAF) of thousands of peacekeeping troops
to enter Somalia so that relief supplies could be safely delivered. ‗Operation Restore Hope‘, as the
mission was called, was the first peacekeeping operation which intervened in a sovereign member state
when that state did not present a military threat to its neighbors (Makinda, 1993). This move challenged
the sovereignty of states by entering a country in the absence of an external threat. Somalian society was
portrayed as an anarchy imploding on itself, and humanitarian need was considered grave enough to
warrant multilateral intervention (Shohat & Stam, 1994). Operation Restore Hope provided the
peacekeepers and military presence necessary for the establishment of the preventive zone.14
Caught between the threats of President Moi and the safety of the refugees, the Cross-Border Operation
was a response to these political realities. At the request of the UN Secretary-General, UNHCR initiated
the Cross Border Operation based on the concept of a preventive zone, previously used in Afghanistan
and Cambodia (Kirkby, 1997). Its purpose was to stem the flow of refugees from Somalia to Kenya and
to entice those refugees already in Kenya to come home.
(overhead: note the arrows of ‗wishful thinking‘ on the part of UNHCR who used ‗intentional mapping‘
(Harley, 1992) to illustrate their humanitarian aspirations)
Four UNHCR ‗outposts‘, administered from Nairobi, were established at various locations roughly one
hundred miles from the Kenya-Somalia border, as part of the Cross Border Operation. The philosophy
of the outposts was clear: presence equals protection. By establishing UNHCR offices and providing
international civilian staff to complement the UNITAF peacekeepers, the UN believed that it created a
safe area. As one of the people assigned to such an outpost and paid $20 per day in ‗danger pay,‘ I was
less certain. I was responsible for the repatriation of refugees from Kenya in an effective and timely
manner, and yet it became clear that Southern Somalia was not simply the safely patrolled place and
‗preventive zone‘ that it had been designated. Our mere presence did little to convince Somalian
refugees in Kenya that it was safe to return. By June 1993, some 12,000 Somalian refugees had returned
with the help of UNHCR, but the vast majority remained in Kenyan camps (UNHCR, 1993).
UNITAF forces were replaced by UN forces in May 1993; the name of the UN force was UNOSOM II.
Protection has a much different meaning and set of associated practices in the Somalian context. The
presence of international staff, such as myself, in fact jeopardized our protection and that of our vehicles
when, for example, we drove them ourselves. Every Somalian belongs to a politicized economy of
power based on clan membership and location. To be outside this economy of power, as we expatriates
were when we drove UNHCR vehicles rather than let a Somalian UNHCR staff member Somalia drive,
was to be outside the boundaries of patrolled behavior. No number of peacekeepers or UNHCR staff
could retrieve a stolen vehicle or a kidnapped relief worker on their own, but working with the
appropriate clan elders who could discipline wayward members of their own clan, vehicles were
recovered and incidents prevented.
This geographically and culturally inscribed notion of security was further illustrated in May 1993 when,
as the temporary officer in charge of the UNHCR outpost in Bardera, a convoy of returning refugees
from Kenya was ambushed by armed locals along the route. We lost radio contact with the UNHCR
expatriate and local staff in the vehicle escorting the convoy, indicating any number of potentially
negative outcomes. Although we had ready access to armed medical evacuation helicopters staffed by
US army pilots, I opted to wait out the incident. Eventually we re-established radio contact and
discovered that the so-called bandits who had attacked the convoy were, in fact, recognized by refugees
in the convoy as being members of the same clan. The refugees in the convoy reportedly chastised the
young bandits whose aim was not to harm the returning refugees at all, but to steal the UNHCR sport
utility vehicle and radio set. Protection in this context had little to do with UNHCR escorts or a
peacekeeping presence. It was safe for refugees to return because they were coming back to a region of
Somalia controlled by their own clan.
Humanitarian operations in both Somalia and Bosnia deployed peacekeepers to ensure the delivery of
humanitarian aid. This served to blur the distinction between military and humanitarian operations,
something which contributed to my own decision to leave the UN in Somalia. On June 5, 1993, a UN
vehicle carrying 14 peacekeepers was ambushed by the forces of Mohammed Farah Aideed in
Mogadishu, the Somalian capital. When US Cobra attack helicopters then began shooting on
Mogadishu, in an effort to find or kill Mohammed Farah Aideed, no UN employee — whether
peacekeeper or aid worker — was safe. Nor was the notion of ‗peacekeeping‘ any longer applicable. UN
forces killed Somalian civilians in the attacks, provoking threats of revenge from Aideed supporters.
Confrontation between UN and Aideed forces in Mogadishu damaged the UN reputation of neutrality
and hastened the peacekeepers‘ departure. By the end of March 1995, almost all peacekeepers had left
Somalia, leaving more than 150,000 Somali refugees in Kenyan camps. While UNHCR designated and
mapped the preventive zone as a safe space, the majority of Somali refugees in Kenya stayed put. As in
Srebrenica, there was a crucial difference between UN intent and people‘s perceptions of safety on the
Was Northern Iraq‘s Operation Provide Comfort in 1991 truly the first UN-protected area? As a UN
Security Council-sanctioned space, guarded with UN peacekeeping troops, it was. But a lesser known
UNHCR operation in Northern Sri Lanka was quietly making history as a successful sanctuary from war
in that country as early as 1990. UNHCR established an Open Relief Centre (ORC) at an existing
Catholic church and sanctuary at Madhu, in the Mannar District, that year (UNHCR, 1999). The ORC
was declared a ―neutral peace zone‖ by UNHCR, and was officially administered by the church under
UNHCR‘s supervision (Jeyaraj, 1999). The ORC was a temporary place where displaced people on the
move could ―freely enter or leave and obtain essential relief assistance in a relatively safe environment‖
(Athas, 1999). Like safe areas in Bosnia and Somalia, the ORC existed an island of civilian safety
amidst war. Unlike the protected areas of Bosnia and Somalia, the open relief centre at Madhu was
politically neutral, completely demilitarized, and without an international peacekeeping force.
The population of the original ORC has fluctuated from a peak of 35,000 in the early 1990s to low of
about 5,000 in 1993. As of March 1999, the ORC at Madhu hosted some 10,000 refugees.15 What
distinguishes the ORC from the safe cities of Bosnia and the preventive zone of Somalia is its local
identity as a safe area before UNHCR arrived. The Catholic shrine and church at Madhu is the most
sacred in the country (Perera, 1998). In 1544, Catholic converts from Mannar Island fled to the Madhu
area, fleeing the lethal power of a Hindu king in Jaffna who massacred some 600 Catholics. Only a small
group, who escaped with a wooden statue of the Virgin Mary, made it safely to Madhu. In the eyes of the
survivors, the statue had saved their lives. They erected a church at Madhu, which has become a sacred
site of pilgrimage for Catholics of all ethnic backgrounds ever since (Perera, 1998). In 1982, 250,000
Catholic Sinhalese and 150,000 Catholic Tamils made the pilgrimage to Madhu for the annual festivals;
―Linguistic identities of Sinhala and Tamils were overwhelmed by the over arching feeling of being
Christian and Madhu was one place where the ethnic divide was virtually non-existent‖ (Jeyaraj, 1999).
When UNHCR landed in Madhu and established a ‗zone of peace,‘ they did so with a tradition of 450
years of sanctuary behind them.
The Church allowed no arms on the premises of the 400 acre site. The Liberation Tigers of Tamil Eelam
(LTTE), the militant rebel group fighting for an independent Tamil state in Sri Lanka who controlled the
immediate area surrounding Madhu, obliged the Church by agreeing not to recruit members for its cause
on the premises. Initially, the LTTE attempted to bear arms in the camp, tax the people to raise funds for
its cause, and search for recruits among the population, but ―patiently and quietly the Catholic authorities
negotiated with them‖ (Perera, 1999). For almost ten years, thousands of refugees of different religious
backgrounds lived without incident in the ORC, as the conflict in Sri Lanka waxed and waned.
The Catholic converts escaping Hindu rule in the 16th century bespeaks the colonial influence of the
Portuguese who were followed by successive waves of Dutch and British colonization in Sri Lanka.
Political independence in 1948 marks the debut of self-governance in Sri Lanka after British rule. The
rise of Buddhist nationalism in the 1950s was punctuated, in 1956, with the eruption of communal
rioting between Tamil and Sinhala-speaking groups in the eastern part Sri Lanka. As the state sponsored
a nationalism for the Sinhala-speaking majority, the 1970s witnessed the emergence of separatist groups
espousing a Tamil nationalism, including the Liberation Tigers of Tamil Eelam (LTTE). In 1983,
tensions escalated to widespread and systematic attacks on Tamils throughout the country. Thousands
were killed in Colombo and the South after government (Sinhala) soldiers were massacred by Tamil
forces in the North (Pfaffenberger, 1994).
The war in Sri Lanka did generate a role for peacekeepers, albeit not an internationally-sanctioned one.
In 1987, after an accord was reached between the Indian and Sri Lankan governments (one which
excluded Tamil political leaders and militant groups like the LTTE), Indian peacekeepers arrived in
Northern Sri Lanka to monitor peace and to disarm Tamil militants. ―In theory, the Accord was meant to
solve the ethnic conflict. In practice, it was meant to establish Indian hegemony over Sri Lanka‖
(Institute of Agriculture and Women in Development, 1995: 55). The introduction of Indian
In 1992-93 more Open Relief Centres north of Madhu were established to house returning refugees from India. During
relatively calm periods during the war, residents returned home to the areas they come from, including Jaffna (Athas, 1999;
peacekeepers proved an unpopular move on both sides of the ethnic divide. While the ORC at Madhu
was unaffected, fighting ensued between Indian peacekeeping forces and the LTTE until 1990 when the
peacekeepers withdrew (ibid.).
After almost a decade a ‗zone of peace,‘ the open relief centre at Madhu was captured by the Sri Lankan
military forces during Operation Rana Gosa (Battle Cry) II on March 22, 1999. The government declared
that it had ‗liberated‘ Madhu, though Madhu was never occupied by the LTTE, so this was technically
untrue. ―Like the Vatican in Italy Madhu Church premises were an oasis of Independence in a desert of
LTTE control‖ (Jeyeraj, 1999). The ORC was transformed from a demilitarized neutral space
surrounded by LTTE-controlled territory to an armed site under Sri Lankan Army control. By occupying
the church grounds, many feared that it would become a target of LTTE attack, thereby increasing
insecurity rather than ensuring it. When church officials objected to uniformed soldiers on the premises,
it was explained that they had no change of gear. Later when the Bishop of Mannar complained to Sri
Lankan president, Chandrika Kumaratunga, she assured him that their presence was only temporary (The
Sunday Leader, 1999). They were still there on May 3rd when the Defence Minister came to Madhu and
declared that it had to be emptied by the end of the month (UNHCR, 1999). There was no consultation
with UNHCR, but the officer in charge quickly realized that the decision was ―irreversible.‖ Madhu was
emptied in less than three weeks.16 By the end of May, some 4000 displaced persons were shipped to
Jaffna and the remainder relocated to other facilities for the displaced. Although UNHCR reported that
interview with UNHCR, 1999).
This clearing of refugees was understood by many political analysts and media commentators as a political move. The
government could be looked upon favorably in provincial elections held in April after the gains of Operation Rana Gosa II.
As well, Catholics in the south of Sri Lanka could look forward to the possibility of attending the annual festivals once again,
which could help the government get re-elected in December, which it was.
houses were ―disassembled‖ after the refugees left, my eyewitness report in June, 1999 and those of
media commentators confirm that many homes were burnt and most were missing their roofs.17
After the so-called liberation of Madhu and before their expulsion by the SLA, the situation for
displaced persons [referred to in Sri Lanka as refugees]18 on the ground became much more difficult. In
interviews with UNHCR and Save the Children (1999), an NGO also operating in Madhu, staff at both
locations felt that the conditions for ‗refugees‘ in Madhu declined after the Sri Lankan army took
Under LTTE authority, people in Madhu had free mobility on the uncleared [LTTE-controlled] side. They engaged
in fishing, casual labour, and so on, which was necessary given the embargo [placed on the importation of goods to
the LTTE areas]. This is no longer possible for people who now find themselves in the cleared [army-controlled]
areas with restricted mobility and livelihoods…. Now the embargo is down and goods are available but there is no
work and therefore no money to buy accessible goods; a vacuum has been created. There is no market in Madhu any
longer [formerly one could buy seafood, squid, wild boar, etc.]; there were tailors, sari shops, hairdressers, and food
boutiques… (interview with UNHCR, 1999).
Basically, ‗refugees‘ who were involved in an informal economy of day labour could no longer cross the
front-line of the war to work for the day on the LTTE-controlled side and return to the army-controlled
Madhu at night. The war literally cut them off from their modest means of subsistence.19 The local
NGOs operating in Madhu, under the umbrella of the LTTE, also closed down, shrinking the base of
rehabilitation and income generation projects in the area.
It was mentioned to me that some people may have taken their roofing materials with them, but given the boat trip back to
Jaffna for many, and further dislocation to unknown destinations for many others, this seems unlikely.
The appellation ‗refugee‘ normally refers to someone outsider her/his country of origin who cannot or will not return
because of a fear of persecution. In Sri Lanka, most internally displaced persons are referred to as refugees, in part because
many of them returned from India as refugees in the early 1990s. The terminology used here may be technically incorrect, but
it is the local parlance.
I saw how much of this business had also closed in Thadchanamaruthamadu, another ORC some five km north of Madhu,
when Jaffna Tamils decided to leave the camp on the ship, Lanka Muditha, hired to return refugees to Jaffna, many of whom
were shop owners.
While the ORC at Madhu ceases to exist as a refuge for persons displaced from the war in Northern Sri
Lanka, Madhu‘s tradition as a sanctuary and sacred site of pilgrimage continues. In contrast to
Srebrenica or Zepa, the ORC operated safely for nine years before government-backed forces occupied
the area in March 1999. Unlike the preventive zone in Somalia, this safe space was effective and met its
objectives. Why? The ORC was successful because it built upon a local history of sanctuary and safety.
The meaning and identity of Madhu as a place have evolved over centuries, and both are local. Despite
being recognized internationally and hosting a UNHCR ORC, Madhu was first a sanctuary and church
with local connotations. No force was required to instill such meanings; it was only force that removed
the people who enjoyed this safety.
Concluding Remarks: Whose Safety?
This paper has discussed the antecedents and outcomes of safe areas in three war zones. I have not
addressed state-sponsored efforts to create ‗humanitarian zones‘ in Kosovo, ‗safe towns‘ in Chechnya,
nor regroupment camps in Burundi where the state is a party to the conflict and has a strong interest in
controlling the movements of marginalized or minority citizen groups (York, 1999;The New York Times,
2000b). But it is interesting to take notice of these efforts to mimic the international humanitarian
discourse, even if conflicts of interest are clear: the majority of these protected areas — whether
designated by multilateral organizations like the UN or national governments at war with certain
segments of their own population — are highly politicized zones where declared meaning and actual
outcome can be two very different things.
Safe spaces sponsored by the UN Security Council have met with mixed results. The provision of
UNITAF troops to Somalia in December 1992 provided a sufficient aura of protection to allow the UN
Secretary General to ask UNHCR to create the preventive zone in Southern Somalia. The good
intentions of UN Security Council resolution 824 in May 1993 generated the safe city of Srebrenica. In
neither case, were the places designated nor the parties to the conflict consulted. Whereas the preventive
zone in Somalia was a largely ineffectual political palliative, the UN-protected area of Srebrenica was
punitive for the displaced persons who sought refuge there. Preventive, palliative, and punitive, the safe
havens issued by the UN Security Council were both succeeded in saving lives and failed to protect all
that they had promised.
The rules of war and the safe spaces they create, as inscribed in humanitarian law, are less and less
relevant as warring factions ignore these consensual arrangements. Dutch peacekeepers in Srebrenica
assisted Bosnia Serbs in creating lists of men between 16 and 25, believing Serb claims that they would
be questioned as prisoners of war in accordance with the Geneva Conventions (major instruments of
humanitarian law). In the end, these lists were used to systematize the murder and ethnic cleansing of
between seven and eight thousand Bosnia Muslim men (Leopold, 1999). The Geneva Conventions focus
on the treatment of combatants in war when they are wounded or captured by enemy forces. Yet,
increasingly those adversely affected by war are civilians. At the beginning of the twentieth century,
civilians accounted for 15% of war casualties; today the figure is 90%. Humanitarian law and its
provisions may create some useful protected spaces for combatants, but they are by no means sufficient.
While international humanitarian law is built upon consent in the agreements warring factions forge,
there is another distinct scale and meaning of consent in the context of conflict. Consent is not only
contractual, in the legal sense, but it is historically and geographically constituted through structures of
civil society. The history of Madhu as a sanctuary and neutral space continues today. While the
UNHCR-sponsored open relief centre has been dismantled due to the political exigencies of elections
and the continuing war, the displaced people who lived there until May 1999 enjoyed relative safety and
freedom of movement during their stay. Their removal by the Sri Lankan Army underscores the fact that
all safe spaces established to protect civilians are politicized; none is completely neutral. Although the
removal of the displacees from their refuge and livelihoods could be construed as yet another round of
violence, the process took place in an orderly fashion, without incident. Before the war in Sri Lanka,
Madhu was a zone of peace, and this tradition continues despite the recent forced migration of its
residents. The ORC was constituted in a geographical location inscribed with local and historical
connotations of peace. The ORC built upon this long history of sanctuary in Madhu, extending its
meaning from a local and national scale to an international one.
What are the conditions of possibility for a spatialized notion of protection?
Bruderlein, C. (1999) Towards a New Strategic Approach to Humanitarian Protection and the Use of
Protected Areas, final report of the consultation for Office for the Coordination of Humanitarian Affairs
(OCHA) and Harvard Center for Population and Development Studies, Cambridge MA, May 6.
Campbell, D. (1999) Apartheid Cartography: the political anthropology and spatial effects of
international diplomacy in Bosnia, Political Geography, vol. 18, no. 4, pp. 395-436.
Crossette, B. (1999)
Cunliffe, S. A. and M. Pugh (1997)
Cutts, M. (1999)
Department of Foreign Affairs and International Trade, Canada (1999)
Frelick, B. (1999)
Glenny, M. (1992)
Harley, B. (1992) Rereading the Maps of the Columbian Encounter, Annals of the Assoc. of American
Geographers 82(3), pp. 522-542.
Hathaway, J.C. (1994)
HINA news agency (1998)
Hyndman, J. (2000)
Institute of Agriculture and Women in Development (1995)
Jeyaraj, D.B.S. (1999)
Kirkby, J. (1997)
Leopold, E. (1999) UN: World failed to confront evils in Srebrenica, Reuters, November 24.
Makinda, S. (1993)
The New York Times (2000) February 5.
(1999) December 2.
Newland, K. (1999)
OCHA (1999) Report on the Inter-agency Expert Consultation on Protected Areas, rapporteur F. Johns,
Harvard University, Cambridge MA, April 7.
Owen, D. (1995)
Perera, J. (1999)
Pfaffenberger, B. (1994) ―Introduction: The Sri Lankan Tamils‖, (eds.) C. Manogaran & B.
Pfaffenberger, The Sri Lankan Tamils: Ethnicity and Identity. (Boulder/Oxford: Westview Press), pp. 1-
Power, S. (1995)
Save the Children (1999) Personal interview with Country Program Coordinator, June 8.
Shohat, E. and R. Stam (1994)
Sparke, M. (1995) Between Demythologizing and Deconstructing the Map: Shawnadithit‘s New-found-
land and the Alienation of Canada, Cartographica, vol. 32, no. 1, pp. 1-21.
The Sunday Leader (1999)
UNHCR (1999) Personal interview with officer-in-charge for Madhu, June 8.